Finding process parameters for laser-drilled blind holes often relies on an engineer’s experience and the trial-and-error method. However, determining such parameters should be possible using methodical calculations. Studies have already begun to examine the use of neural networks to improve the efficiency of this situation. This study extends the field of research by applying artificial neural networks (ANNs) to predict the required parameters for drilling stainless steel with a certain depth and diameter of blind holes, and it also pre-simulates the drilling result of these predicted parameters before actual laser processing. The pre-simulated drilling results were also compared with real-world observations after drilling the stainless steel. These experimental findings confirmed that the proposed method can be used to accurately select laser drilling parameters and predict results in advance. Being able to make these predictions successfully reduces time spent, manpower, and the number of trial-and-error shots required in the pre-processing phase. In addition to providing specific data for engineers to use, this method could also be used to develop a set of reference parameters, greatly simplifying the laser drilling process.

The wire bonding process is one of the most critical processes in semiconductor packaging. The electrical performance and reliability of IC chips must be evaluated by probe testing before wire bonding. In the probe test, the probe usually leaves traces on the surface of the pad. Large probe marks tend to cause a decrease in bond adhesion and so increase the possibility of ball bond lifting from the pads. Shear force is an important measure of the adhesion between the ball and pad. The prediction of the shear force helps to understand the bonding quality in advance. In this study, the six features of the probe marks were extracted by the automatic image recognition method. Using three machine learning techniques (logistic regression, support vector machine, and random forest) based on principal component analysis (PCA), the shear force is estimated based on six features before wire bonding. The results indicate that the proposed PCA-based random forest could identify bad chips with 97.92% accuracy before wire bonding. Unnecessary chips can be found early, and unnecessary wire bonding can be avoided, thereby saving process time and cost. The proposed method would improve the efficiency and quality yield of the semiconductor packaging process.

This article presents a digital image processing approach applied to estimate the probe mark area on semiconductor wafer pads. The value of mark area can be used to judge the acceptance of the wafer. Before wire bonding to the wafer, a probing needle contacts each pad to test the electrical characteristics of the chip. However, this contact leaves probe marks on the pad. A large probe mark area results in poor adhesion forces at the bond ball of the pad, thus leading to undesirable products. Traditionally, given the difficulty of calculating the area of the irregular probe mark, probe mark area calculations were substituted by calculating the area of the oval that is manually drawn to cover the probe mark area. Nevertheless, this method is inaccurate, and the results varied from person to person. In this article, we present an imaging processing approach to calculate the probe mark area utilizing high-magnification microscopes to capture probe mark images. Our approach is faster and more accurate compared to the traditional method.